Merge pull request #16 from helsingborg-stad/fix/malformed-utf8-esape

Fix/malformed utf8 esape

source/php/Index.Test.php

@@ -160,7 +160,19 @@ public function testThatABlockPostIsGeneratingAExcerpt() {
     $this->assertNotEmpty($truncatedExcerpt);
   }
 
-}
+  public function testThatMalformedUTF8ContentIsFixed() { 
+
+    // Given
+    $post = [
+      "post_title" => "Test Post",
+      "post_content" => "R\xc3\xb8d P\xc3\xb8lse 🌭"
+    ];
 
+    // When
+    $post = Index::utf8ize($post);
 
+    // Then
+    $this->assertEquals("Rød Pølse 🌭", $post['post_content']);
+  }
 
+}

source/php/Index.php

@@ -132,6 +132,8 @@ public function index($post)
             }
         });
 
+        $post = self::utf8ize($post); // UTF-8 Escape
+
         //Index post
         if (self::recordToLarge($post)) {
             $splitRecord = self::splitRecord($post);
@@ -462,4 +464,25 @@ private static function getPostAndPostId($post)
 
         return [$post, $postId];
     }
+
+    /**
+     * Convert data to utf-8
+     *
+     * @param mixed $data
+     * @return mixed
+     */
+    public static function utf8ize($data) {
+        if (is_array($data)) {
+            foreach ($data as $key => $value) {
+                $data[$key] = self::utf8ize($value);
+            }
+        } else if (is_object($data)) {
+            foreach ($data as $key => $value) {
+                $data->$key = self::utf8ize($value);
+            }
+        } else if (is_string($data)) {
+            return mb_convert_encoding($data, 'UTF-8');
+        }
+        return $data;
+    }
 }